School bullying is a type of aggressive behavior that involves an imbalance of power, whereby aggressors deliberately and repeatedly torment or harass a less-dominant person or people to cause physical or mental discomfort, or even injury (Olweus, 2005). The nature of school bullying is an alienated social interaction between students, in which the powerful side suppresses the weak side, creating an oppressive relationship (Luo, 2023). Students in rural western China, who are in the midst of dramatic social changes, such as the disintegration of traditional social structure and the anomie of social rules (Song & Tian, 2011), are experiencing multiple disadvantages including the breakdown of family structures and a lack of educational resources in schools (Yang, 2019). Their families, villages, and schools are governed by a combination of conventions, such as grandparents providing only basic needs for a child and not education; adults not interfering with children’s interactions as the inappropriate words and deeds of other people’s children do not concern them; and schools not reporting incidents to superiors.  These structural problems allow for the collective absence of educational participants in the prevention and mitigation of school bullying (Yang, 2019), exposing students to more bullying risks in schools (Hu & Li, 2018). Therefore, research into bullying among students in rural western China had great practical significance, from which an in-depth analysis of the generating logic of school bullying can be realized.

 

Social-emotional competence (SEC) is a set of core competencies related to individual adjustment and social development that is acquired and applied in the complex contexts of growth and development (Osher et al., 2016). The core of SEC lies in helping individuals reshape their relationships with others and themselves, thus constructing an interactive relationship between individuals and others (Du & Mao, 2018). SEC is a relational social construction that promotes the construction of positive relationships with others based on emotional awareness and management (Du & Mao, 2018). Previous studies have already shown that SEC can improve academic performance and well-being (Durlak et al., 2011), facilitate adjustment to school life (Shields et al., 2001), and reduce bullying behaviors among adolescents (Wilson et al., 2001). Although these studies have found a significant relationship between SEC and school bullying, they are not without limitations. First, the studies mainly focused on exploring the unidirectional predictive relationship of SEC on school bullying, neglecting the reverse effect of the latter variable on the former. Second, most studies adopted cross-sectional data to reveal the impact of SEC on school bullying; therefore, the long-term effect of this relationship remains to be seen due to a lack of dynamic tracking. Third, students have generally been studied as a homogenized group in previous studies; therefore, the heterogeneity of the relationship between SEC and school bullying in different student groups has also been neglected.

 

To address these limitations, this study explored the bidirectional relationship, immediate effect, and longitudinal effect of SEC on school bullying by constructing a cross-lagged model through a 2-year longitudinal study of rural students in China.

SEC, as the competence of individuals to manage their emotions as well as to construct social relationships, plays a vital role in students dealing with school bullying and coping with the negative emotions (Malecki et al., 2015). According to interaction theory, emotion is a behavioral regulator, affecting individuals’ behavior and interactions, and influencing their long-term behaviors (Collins, 2012). When individuals experience school bullying, they manage their emotions by adjusting their behavior and choosing effective ways to interact, so as to avoid being hurt (He et al., 2023).

 

Previous studies have found that higher noncognitive competence of SEC can reduce students’ bullying behaviors and other antisocial behaviors (He et al., 2023; Wilson et al., 2001). Adolescents with lower SEC tend to exhibit lower self-perception and self-evaluation, continually signaling their inability to protect themselves from bullies (Malecki et al., 2015). Students with high SEC tend to employ positive and effective strategies to cope with bullying, such as rejecting the bully’s imposition of their aggressive overtures (Mahady Wilton et al., 2000). In addition, increased self-management of SEC can reduce problem behaviors (Jones et al., 2014). Thus, this study posited the following hypothesis:

Hypothesis 1: The level of students’ social-emotional competence will significantly and negatively predict the degree to which students experience bullying.

The negative emotional experiences, such as panic and loneliness, that result from being bullied at school can affect individuals’ identification of, encoding of, interpretation of, and emotional responses to emotional cues, resulting in emotional disorders (Bond et al., 2001; Troop‐Gordon & Ladd, 2005). This, in turn, negatively impacts the emotional cognition and management competence of the bullied. According to the general aggression model, the proximate process often causes an immediate assessment of the situation, which occurs automatically (i.e., spontaneously and unconsciously) and is influenced by the person’s current internal state (Allen et al., 2018). During the initial stages of a student’s exposure to school bullying, negative emotions such as frustration and depression often occur instantly and in an out-of-control manner (Troop‐Gordon & Ladd, 2005), which negatively affects SEC (He et al., 2023).

 

At the same time, victims are prone to feel isolated and are unable to improve their social relationships, causing them to experience socialization problems (e.g., interpersonal disorders) and problem-solving biases (Perren et al., 2013). This shows a negative impact on students’ SEC, and even makes their SEC retrogressive. For example, one study found that prolonged exposure to bullying in schools led to the victims lacking a sense of security and a sense of belonging in their schools (Xu & Fang, 2021). Taking a negative attitude toward their schools tended to reduce their emotional connection with the school community, and trigger externalizing problem behaviors such as aggression, delinquency, and truancy (Cross et al., 2015). Furthermore, O’Brennan et al. (2009) argued that bullying poses a threat to children’s social-emotional functioning. In their analysis of data from 24,345 school students, they found that bullies and victims exhibited internalizing symptoms and peer relationship problems (O'Brennan et al., 2009). Thus, this study suggested the following hypothesis:

Hypothesis 2: Bullying will significantly and negatively predict the level of adolescents’ social-emotional competence.

Children in western China typically have one of two types of background: left behind and boarding. Left-behind children live with their grandparents while their parents go to work far away. Boarding children live in dormitories at a boarding school. Left-behind children and boarding children are two different types of children who do not live with their parents.

 

From the perspective of social interaction, both SEC and school bullying are socialized interaction processes in complex environments. In rural areas of western China, left-behind children and boarding students are two widely distributed social groups that face developmental dilemmas such as parent–child isolation, lack of family education and social education, and suffering from serious bullying (Wang et al., 2017).

 

Young et al. (2013) found for left-behind children, the long-term absence of parents during their emotional development process may weaken their positive perception of social support systems including family, school, and society. This may negatively impact their development of SEC, which is closely connected with their bullying behaviors. An empirical study of 112 schools from 28 counties in China showed that parental absence was negatively associated with the development of left-behind children (Mao et al., 2020). When left-behind children’s socioemotional development is impaired, they are more likely to suffer from negative emotions that are related to bullying or being bullied, such as loneliness, depression, or anger (Dai & Chu, 2018). Another study showed that the left-behind experience was an important risk factor for mental health problems among students, with a prevalence of mental health problems of 35.69% among left-behind students compared to 19.68% among students without left-behind experience (H. Liu et al., 2020). In view of the above analysis, we proposed the following hypothesis:

Hypothesis 3a: The negative predictive effect between being bullied and social-emotional competence will be stronger for left-behind children than for non-left-behind children.

 

There are many boarding students in rural western China. Due to facility insufficiency, caregiver shortages, and poor services (Wang et al., 2017), these boarders may not receive compensatory emotional support, making them prone to suffer from more emotional and interpersonal barriers, and more bullying than that experienced by nonboarding students (Pfeiffer & Pinquart, 2014). In other words, when students’ social-emotional competence level is too low to allow them to cope with the danger of school bullying, boarders (vs. nonboarders) have more risk factors and are more likely to be persecuted by school bullies because they have no other compensatory emotional support in school. A lack of family emotional support means these problems exert a significant negative impact on their SEC (Pfeiffer & Pinquart, 2014). One study found that boarding life is not a stable protective factor for the development of students’ social-emotional competence but it does have a negative impact on the development of social-emotional competence (Wang & Mao, 2015). That is, boarders who are persecuted by school bullying are not protected by boarding life, which strengthens the negative impact on the development of social-emotional competence. Therefore, we proposed the following hypothesis:   

Hypothesis 3b: The negative predictive effect between school bullying and social-emotional competence will be stronger for boarding children than for nonboarding children.

 

The proposed cross-lagged model is shown in Figure 1.

 

We selected 21 rural primary and secondary schools from two counties in western China, and conducted initial and follow-up surveys at baseline and 2 years later using the cluster sampling method. The first survey (T1) was conducted in September 2018 and we collected 2,271 forms. After invalid surveys such as those with high answer repetition rates or too many missing values, or that had not been filled in correctly were deleted, 2,065 valid questionnaires were retained, for an effective rate of response of 90.9%. Of these, 888 (43.0%) were from fifth graders, 724 (35.1%) from sixth graders, and 453 (21.9%) from seventh graders. There were 1,069 (51.8%) girls, 960 (46.5%) boys, and 36 (1.7%) did not specify their gender.

 

The second questionnaire survey (T2) was conducted in December 2020. After collecting 1,629 surveys and deleting invalid forms, 1,452 valid surveys were retained, for an effective rate of response of 89.1%. Of these, 596 (41.0%) were from seventh graders, 488 (33.6%) from eighth graders, and 368 (25.3%) from ninth graders. There were 778 (53.6%) girls, 667 (45.9%) boys, and seven (0.5%) who did not specify their gender. Comparison between the second survey (T2) and the first survey (T1) showed a sample loss of 613, signifying a longitudinal attrition rate of 29.7%. After conducting an independent samples t test between the attrition sample at T1 and the valid sample in terms of the core variables of school bullying and SEC, we found that no significant differences between either school bullying, t = –1.03, p = .30, or SEC, t = 0.63, p = .53, suggesting that there were no structural deficiencies in the valid samples that completed both surveys.

We measured SEC with a self-assessed questionnaire comprising 30 items divided across six dimensions: self-awareness, self-management, other-awareness, other-management, collective-awareness, and collective-management (Y. Chen & Mao, 2016). The scale is scored on a 5-point Likert scale ranging from 1 (completely disagree) to 5 (completely agree), with higher scores representing higher levels of development of students’ SEC. Sample items are “I know what I do well” (self-awareness dimension), “I can learn experience from others’ successes” (self-management dimension), “I understand the choices other people make” (other-awareness dimension), “I get along well with my classmates” (other-management dimension), “I am happy to be a member of the class” (collective-awareness dimension), and “In group activities I actively talk with my peers” (collective-management dimension). Cronbach’s alpha values for the scale were .87 (T1) and .92 (T2) in this study, indicating it was reliable. There was also good reliability and validity, T1: χ²/df = 3.74, comparative fit index (CFI) = .91, root-mean-square error of approximation (RMSEA) = .04, Tucker–Lewis index (TLI) = .90, goodness-of-fit index (GFI) = .95; T2: χ²/df = 5.34, CFI = .88, RMSEA = .06, TLI = .86, GFI = .90.

 

To measure school bullying we used the Chinese version (He et al., 2019) of the Child Development Project questionnaire, which consists of three dimensions: verbal bullying, property bullying, and physical bullying. A sample item is “Did someone ever make fun of you, call you names, or harass you?” Students report the frequency of each type of bullying in the past year on a 4-point Likert scale ranging from 1 (never) to 4 (always). Higher scores indicate more severe levels of perceived bullying. Cronbach’s alpha values for the scale were .78 (T1) and .76 (T2), indicating it was reliable. There was also good reliability and validity, T1: χ²/df = 6.41, CFI = .99, RMSEA = .05, TLI = .97, GFI = .99; T2: χ²/df = 5.92, CFI = .98, RMSEA = .06, TLI = .96, GFI = .99.

We used SPSS 22.0 for data entry, descriptive statistical analysis, correlation analysis, and internal consistency reliability analysis. The students were classified into two groups based on whether they lived in school dormitories. The first group comprised boarding students, which means that students leave home and live in the school dormitory. The second group was nonboarding students, which means that students live at home. Left-behind students were classified into four groups according to whether their parents migrated for work. The first group was double left-behind, which meant that both parents migrated for work. The second group was father away, which meant only the father migrated for work and the mother stayed at home. The third group, whose mothers migrated for work and whose fathers lived at home, was called mother away. The fourth group, with both parents living at home, was the non-left-behind group. We used Amos 17.0 for confirmatory factor analysis, latent variable structural equation model, and multi-group cross-lagged model difference tests.

The problem of common method biases inevitably arises when questionnaires are used to collect data. To reduce common method biases, researchers mostly use procedural and statistical controls. In this study we applied Harman’s single-factor test to examine the severity of common method bias. An unrotated principal component factor analysis revealed seven factors with eigenvalues greater than 1. The first factor explained 20.8% (T1) and 26.2% (T2) of the total variance, both of which are below the critical value of 40%, indicating that there was no significant common method bias in this study.

Table 1 shows the relationship between school bullying and SEC. There were significant positive correlations between SEC at T1 and at T2, and between school bullying at T1 and at T2. Further, at both T1 and at T2 there was a significant negative correlation between SEC and school bullying.

We conducted a 2 (time: T1 vs. T2) × 4 (type of left-behind child: double left-behind child vs. father away vs. mother away vs. non-left-behind child) mixed analysis of variance (ANOVA) with time of measurement as the within-subjects variable, type of left-behind child as the between-subjects variable, and SEC as the dependent variable. Results showed that the main effect of time of measurement was significant, F(1, 1448) = 65.45, p < .001, η_p² = .04, meaning that SEC at T2 (M = 3.97, SD = 0.53)  was significantly higher than that at T1 (M = 3.81, SD = 0.56). The main effect of type of left-behind child was not significant, F(3, 1448) = 1.86, p = .134. The interaction effect between time of measurement and type of left-behind child reached marginal significance, F(3, 1448) = 2.51, p = .057. Further analysis of simple effects revealed that among the different types of left-behind children, there was a significant difference in students’ SEC at T1, F(3, 1448) = 2.66, p = .047, η_p² = .01. To be specific, non-left-behind students (M = 3.91, SD = 0.48) scored significantly higher than those for whom both parents were away (M = 3.88, SD = 0.43, p = .023), or the mother was away (M = 3.81, SD = 0.45, p = .027). The corresponding difference was not significant at T2, F(3, 1448) = 1.43, p = .232.

 

Next, we conducted another 2 (time: T1 vs T2) × 4 (type of left-behind child: double left-behind child vs. father away vs. mother away vs. non-left-behind child) repeated mixed ANOVA, with time of measurement as the within-subjects variable, type of left-behind child as the between-subjects variable, and school bullying as the dependent variable. Results showed that the main effect of time of measurement was significant, F(1, 1448) = 149.57, p < .001, η_p² = .09, meaning that levels of school bullying at T2 (M = 1.47, SD = 0.43) were significantly lower than those at T1 (M = 1.70, SD = 0.61). Further, the main effect of type of left-behind child was significant, F(3, 1448) = 3.64, p = .012, η_p² = .01. Pairwise comparisons revealed that children whose mother was away (M = 1.68, SD = 0.45) experienced significantly higher levels of school bullying than did non-left-behind students (M = 1.56, SD = 0.42, p = .001), double left-behind children (M = 1.58, SD = 0.41, p = .007), and those whose father was away (M = 1.60, SD = 0.41, p = .048). The interaction effect between time of measurement and type of left-behind child was not significant, F(3, 1448) = 1.08, p = .357.

 

Next, we conducted a 2 (time: T1 vs. T2) × 2 (boarding status: boarding vs. non-boarding) mixed ANOVA with time of measurement as the within-subjects variable, boarding status as the between-subjects variable, and SEC as the dependent variable. Results showed that the main effect of time of measurement was significant, F(1, 1434) = 26.07, p < .001, η_p² = .02, meaning that SEC at T2 (M = 3.97, SD = 0.53) was significantly higher than that at T1 (M = 3.81, SD = 0.56). The main effect of boarding status was not significant F(1, 1434) = 0.39, p = .534. The interaction effect between time of measurement and boarding status was not significant, F(1, 1434) = 0.26, p = .613.

 

Finally, we conducted another 2 (time: T1 vs T2) × 2 repeated (boarding status: boarding vs nonboarding) mixed ANOVA with time of measurement as the within-subjects variable, boarding status as the between-subjects variable, and school bullying as the dependent variable. Results showed that the main effect of time of measurement was significant, F(1, 1434) = 40.28, p < .001, η_p² = .03, meaning that levels of school bullying at T2 (M = 1.47, SD = 0.43) were significantly lower than that at T1 (M = 1.70, SD = 0.61). The main effect of boarding status was marginally significant, F(1, 1434) = 3.51, p = .061. To be specific, boarders (M = 1.58, SD = 0.41) experienced lower levels of school bullying than did nonboarders (M = 1.66, SD = 0.45). The interaction effect between time of measurement and boarding status was not significant, F(1, 1434) = 0.001, p = .973.

With reference to van Lier et al.’s (2012) procedure, this study built four models to examine the cross-lagged relationships among the longitudinal tracking data. The fit results for each model are shown in Table 2. Model 1 was the baseline model, with only the autoregressive model of SEC and school bullying. Model 2 included the predictive path of SEC to school bullying being added to Model 1. Model 3 included the predictive path of school bullying on SEC being added to Model 1. Model 4 included the cross-lagged path of SEC and school bullying being added to Model 1. The fit results of the four models showed χ²/df values between 3 and 5, which meant they were loading at an acceptable level; RMSEA < .05, CFI > .90, TLI > .90, and GFI > .90. Thus, all four models demonstrated a high degree of fit to the data.

 

We further determined the best option among the four models proposed by calculating the Brwone-Cudeck criterion (BCC) and Bayes information criterion (BIC; Burnham & Anderson, 1998). We added a constant to all BCCs to make the smallest BCC become 0 and thus yield BCC₀, and added a constant to all BICs to make the smallest BIC become 0, thus yielding BIC₀. Both the BCC₀ and BIC₀ of Model 4 were 0, and the BCC₀ and BIC₀ of Models 1, 2, and 3 were all greater than 2, so Model 4 was the best option. We found that Model 4 had a better fit compared to other models, so we used this cross-lagged model to examine the longitudinal relationship between socioemotional competence and school bullying (see Table 2).

Results of the cross-lagged model (Model 4) analysis were as follows: First, SEC at T1 significantly and positively predicted SEC at T2, β = .35, p < .001. Second, school bullying at T1 significantly and positively predicted school bullying at T2, β = .29, p < .001. Third, SEC at T1 significantly and negatively predicted school bullying at T2, β = –0.20, p < .001. Fourth, school bullying at T1 significantly and negatively predicted SEC at T2, β = –.10, p < .001. This further validates that there was a bidirectional predictive relationship between school bullying and SEC, and the relationship exhibited a longitudinal nature; that is, SEC had a longitudinal protective effect on school bullying, while school bullying had a longitudinal negative effect on SEC (see Figure 2). The effect was significant; thus, Hypotheses 1 and 2 were supported.

 

To inspect the differences in the interrelationship between adolescents’ SEC and school bullying in terms of left-behind status, we used a multigroup comparison method of cross-lagged analysis. First, we set each path coefficient to be freely estimated to obtain the measurement coefficient model (M5–1), then we set the path coefficients of four different groups of retention types to be equal to obtain the equivalent estimation model (M6–1). After that, we compared M5–1 with M6–1. Results showed that M5–1 and M6–1 had good fit level, M5–1: χ² = 1897.22, df = 1044, CFI = .93, TLI = .93, GFI = .90, SRMR = .05, RMSEA = .02; M6–1: χ² = 1913.78, df = 1056, CFI = .93, TLI = .93, GFI = .90, SRMR = .05, RMSEA = .02. However, the difference between the fit results of the two models was not significant, Δχ² = 16.56, Δdf = 12, p = .167, showing that the four types of left-behind children were not significantly different in the structural model coefficients and were invariant between groups. Thus, Hypothesis 3a was not supported.

 

Similarly, we inspected the differences in the interrelationship between adolescents’ SEC and school bullying in terms of boarding status. The free estimation model M5–2 and the equivalent estimation model M6-2 were obtained. The fit indices of M5–2, χ² = 1301.47, df = 512, CFI = .94, TLI = .93, GFI = .93, SRMR = .03, RMSEA = .03, and M6–2, χ² = 1307.05, df = 516, CFI = .94, TLI = .93, GFI = .93, SRMR = .03, RMSEA = .03, showed a good fit to the data, and a comparative analysis revealed that the difference between the fit results of the two models was not significant, Δχ² = 5.58, Δdf = 4, p = .232. Thus, Hypothesis 3b was not supported.

 

We further classified students into four groups: regular students (neither left behind nor boarding), single-boarding students (boarding but not left behind), single-left-behind (left behind but not boarding) students, and left-behind boarders. A difference test of the cross-lagged model was conducted. First, the four path coefficients of cross-lag were set to be freely estimated to obtain model M5–3; then the cross-lagged path coefficients of the four groups of students were set to be equal to obtain model M6–3. A comparison of M5–3 with M6–3 revealed that both had good fit to the data, M5–3: χ² = 2061.17, df = 1044, CFI = .92, TLI = .91, GFI = .90, RMSEA = .03; M6-3: χ² = 2084.37, df = 1056, CFI = .92, TLI = .91, GFI = .90, RMSEA = .03. The difference between the fit results of the two models was significant, Δχ² = 23.20, Δdf = 12, p = .026, indicating that the four groups of students showed significant differences in the structural model coefficients.

 

We also conducted a paired difference test on the individual path coefficients of the four-group cross-lagged model, and found that there were significant differences in the autoregressive path of SEC, the autoregressive path of school bullying, and the path of school bullying predicting SEC. Finally, the paths with significant differences were freely estimated, while other paths with nonsignificant differences were restricted to equalize the coefficients of the four groups, and model M7–3 was obtained. The overall fit of this model was χ² = 2062.95, df = 1047, CFI = .92, TLI = .91, GFI = .90, RMSEA = .03.

 

From the fit indices of M7–3 and the coefficients of each path (see Table 3), we found that from T1 to T2 stability (of different degrees) in SEC was found among regular students, single-boarding students, and left-behind boarders, while excluding single-left-behind students. The stability of school bullying of single-left-behind students was higher than that of other student groups; that is, single-left-behind students who previously suffered from school bullying were more likely to suffer from it again after an interval. Among all student groups, SEC showed a long-term stabilizing protective effect against school bullying. School bullying was negatively predictive of SEC to varying degrees for regular students, single-boarding students, and left-behind boarders, but the effect was not significant for single-left-behind students. Thus, Hypotheses 3a and 3b were not supported by independent tests.

In this study we found a significant main effect of time on SEC, with a significant tendency for SEC to improve after an interval of 2 years. This finding is consistent with those of Hébert et al. (2014). As students’ noncognitive competence develops, the scope of social interactions expands, their experience in social interactions accumulates, and their SEC continues to develop. According to interaction ritual chain theory, an individual’s emotional energy and symbolic reserves can be reproduced, and their emotional capacity can be reshaped in interaction (Collins, 2012). In addition, in this study we found an interaction effect between time of measurement and type of left-behind child that was marginally significant, showing that the SEC of the double left-behind students and those whose mothers left home was significantly lower than that of non-left-behind students at the early stage of the test, while at T2 the difference between the two groups diminished.

 

A repeated-measures ANOVA with school bullying as the dependent variable revealed a significant main effect of time on school bullying, showing a significant tendency to decrease with age. This is consistent with the findings of Pryce and Frederickson (2013). As students grow up, they experience a steady development of SEC, enabling them to better manage their emotions and adopt more positive strategies for social interactions in school bullying. In addition, left-behind and boarding status had significant or borderline significant effects on school bullying, which was stable, showing that students whose mothers left home suffered more severe school bullying, while boarding students suffered less school bullying than did nonboarding students. In rural areas of western China, mothers migrating for work often occurs along with fathers’ functional absence from the family economy. Boarding, on the other hand, as another form of parent–child separation, played a somewhat protective role against school bullying. This finding is not consistent with that of Pfeiffer and Pinquart (2014). This may be because of the joint implementation of policies on school bullying and other issues.

Cross-lagged modeling showed that the initial SEC of students significantly and negatively predicted school bullying after a 2-year interval. This finding breaks through the limitation of existing studies that adopted cross-sectional data for immediate effect analysis and verifies that SEC has a long-lasting protective effect against school bullying. Students with high SEC can better manage their emotions; therefore, school bullying behaviors are reduced. Further, the SEC that has been developed will not fade away, but rather be internalized as students’ favorable emotional qualities and behavioral tendencies, which can protect them from school bullying in the long term.

 

As an additional pathway of the bidirectional relationship, this study showed that school bullying had a long-term negative predictive effect on SEC, further validating the persistent negative impact of school bullying on students’ individual development (Xie et al., 2019). According to the general aggression model, students who suffer from bullying in school tend to feel frustrated, angry, sad, and depressed at the initial stage, and these negative emotions appear instantly along with the bullying incident, showing proximal effects. In addition, if these negative emotions cannot be resolved in time and the bullying cannot be effectively blocked, the bullied student will experience dissipation of intrinsic positive emotions and accumulation of negative emotions (J. Chen et al., 2022). At the same time, school bullying incidents, together with the accompanying negative emotional cues, will be encoded, interpreted, and stored by students (Lemerise & Arsenio, 2000).

We found neither differences between left-behind and non-left-behind adolescents nor differences between boarding and nonboarding students, indicating that the single effect of left-behind or boarding status was not significant. Further comprehensive analysis of left-behind and boarding children revealed that the improvement of SEC in students effectively reduced the level of bullying in school, reflecting the universality and stability of SEC in protecting students against school bullying. In addition, significant variability was demonstrated among different student groups, as evidenced by the nonsignificant autoregressive coefficients of SEC and significant higher autoregressive coefficients of school bullying for single-left-behind students. Moreover, school bullying did not have a significant longitudinal negative predictive effect on SEC for single-left-behind students. In this regard, left-behind but not boarding status was a risk factor for students’ unfavorable development of SEC and exposure to school bullying. This is consistent with previous empirical research, which suggests that boarding can compensate for the negative impact of parental absence on students’ mental health (M. Liu & Villa, 2020). When students are unable (vs. able) to receive stable and consistent social support from school, family and community, their SEC will lose the necessary social context to develop and thus become more (vs. less) contingent and fluctuant. Due to insufficient external support and a weak internal ability (Wang et al., 2017), single-left-behind students are more likely to be labeled as vulnerable groups, thus causing them to become the ideal targets of bullying. Especially for single-left-behind students, their school bullying experience did not show a significant negative predictive effect on SEC. This may be related to their emotional response profile. As multiple disadvantaged students, they are likely to be exposed to more negative life events at school and at home, resulting in a higher threshold for emotional response to stimuli. The likelihood that a particular emotion will be activated depends on the pattern of the stimulus and the individual’s threshold for that emotion, as well as other individual differences (Izard, 1993).

Despite the theoretical and practical implications of this study, there are several limitations. First, due to data limitations, some hypotheses were not well-verified. Future research could further estimate the relationship between bullying and SEC. Second, in terms of sample representation, this study was conducted in the setting of rural schools located in western China, and we are unsure to what extent the findings can be generalized to other districts and other cultural contexts in China. Future studies could explore the applicability of our research model in different cultural and institutional contexts.

In summary, we found a bidirectional, long-term, negative cross-lagged relationship between SEC and school bullying. The cross-lagged models of students’ SEC and school bullying showed significant differences between left-behind and boarding statuses. To avoid bullying, the efficacy of SEC in preventing school bullying should be emphasized, and a comprehensive change model should be conducted to develop students’ SEC and prevent school bullying. Further, educational compensation for children with multiple disadvantages needs to be strengthened to enhance their SEC and prevent school bullying. Sustained attention from the whole of society should be given to socially disadvantaged children in rural areas, especially left-behind and nonboarding children.

Allen, J. J., Anderson, C. A., & Bushman, B. J. (2018). The general aggression model. Current Opinion in Psychology, 19, 75–80. https://doi.org/10.1016/j.copsyc.2017.03.034

 

Bond, L., Carlin, J. B., Thomas, L., Rubin, K., & Patton, G. (2001). Does bullying cause emotional problems? A prospective study of young teenagers. BMJ, 323(7311), 480–484. https://doi.org/10.1136/bmj.323.7311.480

 

Burnham, K. P., & Anderson, D. R. (1998). Model selection and inference. A practical information–theoretic approach. Springer. https://doi.org/10.1007/978-1-4757-2917-7

 

Chen, J., Zhao, Y., Li, K., Tang, C., Yan, M., Zhang, Q., ... Liu, Q. (2022). Analysis of the current situation of cyberbullying among children and adolescents and its correlation with anxiety and depression symptoms [In Chinese]. Modern Preventive Medicine, 49(5), 808–813.

 

Chen, Y., & Mao, Y. (2016). The relationship between family capital and academic achievements of primary school students in western rural areas: The mediating role of social-emotional competence [In Chinese]. Chinese Journal of Special Education, 4, 90–96.

 

Collins, R. (2012). Interactive ritual chains. Commercial Press.

 

Cross, D., Lester, L., & Barnes, A. (2015). A longitudinal study of the social and emotional predictors and consequences of cyber and traditional bullying victimisation. International Journal of Public Health, 60(2), 207–217. https://doi.org/10.1007/s00038-015-0655-1

 

Dai, Q., & Chu, R.-X. (2018). Anxiety, happiness and self-esteem of western Chinese left-behind children. Child Abuse & Neglect, 86, 403–413. https://doi.org/10.1016/j.chiabu.2016.08.002

 

Du, Y., & Mao, Y. (2018). A study on the construction and development of students’ social-emotional competence based on the relationship perspective [In Chinese]. Educational Research, 39(8), 43–50.

 

Durlak, J. A., Weissberg, R. P., Dymnicki, A. B., Taylor, R. D., & Schellinger, K. B. (2011). The impact of enhancing students’ social and emotional learning: A meta‐analysis of school‐based universal interventions. Child Development, 82(1), 405–432. https://doi.org/10.1111/j.1467-8624.2010.01564.x

 

He, E., Ye, X., Pan, K., & Mao, Y. (2019). Impact of elementary school students’ social-emotional learning ability on school bullying: The moderating role of school belonging [In Chinese]. Modern Education Management, 8, 99–105.

 

Hébert, T. P., Corcoran, J. A., Coté, J. M., Ene, M. C., Leighton, E. A., Holmes, A. M., & Padula, D. D. (2014). It’s safe to be smart: Strategies for creating a supportive classroom environment. Gifted Child Today, 37(2), 95–101. https://doi.org/10.1177/1076217514520966

 

Hu, Y., & Li, J. (2018). Who is being bullied?—A study on the influencing factors of school bullying among middle school students [In Chinese]. Journal of Capital Normal University (Social Sciences Edition), 6, 171–185.

 

Izard, C. E. (1993). Four systems for emotion activation: Cognitive and noncognitive processes. Psychological Review, 100(1), 68–90. https://doi.org/10.1037/0033-295x.100.1.68

 

Jones, S. M., Bailey, R., & Jacob, R. (2014). Social-emotional learning is essential to classroom management. Phi Delta Kappan, 96(2), 19–24. https://doi.org/10.1177/0031721714553405

 

Lemerise, E. A., & Arsenio, W. F. (2000). An integrated model of emotion processes and cognition in social information processing. Child Development, 71(1), 107–118. https://doi.org/10.1111/1467-8624.00124

 

Liu, H., Zhou, Z., Fan, X., Wang, J., Sun, H., Shen, C., & Zhai, X. (2020). The influence of left-behind experience on college students’ mental health: A cross-sectional comparative study. International Journal of Environmental Research and Public Health, 17(5), Article 1511. https://doi.org/10.3390/ijerph17051511

 

Liu, M., & Villa, K. M. (2020). Solution or isolation: Is boarding school a good solution for left-behind children in rural China? China Economic Review, 61, Article 101456. https://doi.org/10.1016/j.chieco.2020.101456

 

Luo, C. (2023). A study of school governance of school bullying under the perspective of social interaction [In Chinese]. Educational Theory and Practice, 43(4), 58–64.

 

Mahady Wilton, M. M., Craig, W. M., & Pepler, D. J. (2000). Emotional regulation and display in classroom victims of bullying: Characteristic expressions of affect, coping styles and relevant contextual factors. Social Development, 9(2), 226–245. https://doi.org/10.1111/1467-9507.00121

 

Malecki, C. K., Demaray, M. K., Coyle, S., Geosling, R., Rueger, S. Y., & Becker, L. D. (2015). Frequency, power differential, and intentionality and the relationship to anxiety, depression, and self-esteem for victims of bullying. Child & Youth Care Forum, 44(1), 115–131. https://doi.org/10.1007/s10566-014-9273-y

 

Mao, M., Zang, L., & Zhang, H. (2020). The effects of parental absence on children development: Evidence from left-behind children in China. International Journal of Environmental Research and Public Health, 17(18), Article 6770. https://doi.org/10.3390/ijerph17186770

 

O’Brennan, L. M., Bradshaw, C. P., & Sawyer, A. L. (2009). Examining developmental differences in the social‐emotional problems among frequent bullies, victims, and bully/victims. Psychology in the Schools, 46(2), 100–115. https://doi.org/10.1002/pits.20357

 

Olweus, D. (2005). A useful evaluation design, and effects of the Olweus Bullying Prevention Program. Psychology, Crime and Law, 11(4), 389–402. https://doi.org/10.1080/10683160500255471

 

Osher, D., Kidron, Y., Brackett, M., Dymnicki, A., Jones, S., & Weissberg, R. P. (2016). Advancing the science and practice of social and emotional learning: Looking back and moving forward. Review of Research in Education, 40(1), 644–681. https://doi.org/10.3102/0091732X16673595

 

Perren, S., Ettekal, I., & Ladd, G. (2013). The impact of peer victimization on later maladjustment: Mediating and moderating effects of hostile and self‐blaming attributions. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 54(1), 46–55. https://doi.org/10.1111/j.1469-7610.2012.02618.x

 

Pfeiffer, J. P., & Pinquart, M. (2014). Bullying in German boarding schools: A pilot study. School Psychology International, 35(6), 580–591. https://doi.org/10.1177/0143034314525513

 

Pryce, S., & Frederickson, N. (2013). Bullying behaviour, intentions and classroom ecology. Learning Environments Research, 16, 183–199. https://doi.org/10.1007/s10984-013-9137-7

 

Shields, A., Dickstein, S., Seifer, R., Giusti, L., Dodge Magee, K., & Spritz, B. (2001). Emotional competence and early school adjustment: A study of preschoolers at risk. Early Education and Development, 12(1), 73–96. https://doi.org/10.1207/s15566935eed1201_5

 

Song, L., & Tian, X. (2011). On the circle structure: A reunderstanding of the changes of social structure in contemporary rural China [In Chinese]. Journal of China Agricultural University (Social Sciences Edition), 28(1), 13. 

 

Troop‐Gordon, W., & Ladd, G. W. (2005). Trajectories of peer victimization and perceptions of the self and schoolmates: Precursors to internalizing and externalizing problems. Child Development, 76(5), 1072–1091. https://doi.org/10.1111/j.1467-8624.2005.00898.x

 

van Lier, P. A. C., Vitaro, F., Barker, E. D., Brendgen, M., Tremblay, R. E., & Boivin, M. (2012). Peer victimization, poor academic achievement, and the link between childhood externalizing and internalizing problems. Child Development, 83(5), 1775–1788. https://doi.org/10.1111/j.1467-8624.2012.01802.x

 

Wang, S., Dong, X., & Mao, Y. (2017). The impact of boarding on campus on the social-emotional competence of left-behind children in rural western China. Asia Pacific Education Review, 18(3), 413–423. https://doi.org/10.1007/s12564-017-9476-7

 

Wang, S., & Mao, Y. (2015). The impact of boarding on the development of left-behind children’s social-emotional ability: An empirical study based on 11 provinces in western China [In Chinese]. Journal of Education, 11(5), 111–120.

 

Wilson, D. B., Gottfredson, D. C., & Najaka, S. S. (2001). School-based prevention of problem behaviors: A meta-analysis. Journal of Quantitative Criminology, 17(3), 247–272. https://doi.org/10.1023/A:1011050217296

 

Xie, J., Wei, Y., & Zhuorong, Z. (2019). Exploring patterns of bullying victimization among contemporary Chinese adolescents in schools: A potential profile-based analysis [In Chinese]. Psychological Development and Education, 35(1), 95–102.

 

Xu, Z., & Fang, C. (2021). The relationship between school bullying and subjective well-being: The mediating effect of school belonging. Frontiers in Psychology, 12, Article 725542. https://doi.org/10.3389/fpsyg.2021.725542

 

Yang, H. (2019). Causes and governance of bullying in rural schools under the perspective of lack of educational subject [In Chinese]. Journal of Jishou University (Social Sciences Edition), 40(6), 86–98.

 

Young, B. J., Wallace, D. P., Imig, M., Borgerding, L., Brown-Jacobsen, A. M., & Whiteside, S. P. H. (2013). Parenting behaviors and childhood anxiety: A psychometric investigation of the EMBU-C. Journal of Child and Family Studies, 22(8), 1138–1146. https://doi.org/10.1007/s10826-012-9677-y